Most people access the Fediverse through one of the large instances: lemmy.world, kbin, or beehaw. New or small instances of Lemmy have no content by default, and can most easily get content by linking to larger Lemmy instances. This is done manually one “Community” at a time (I spent 15 minutes doing this yesterday). Meanwhile, on larger instances, content naturally aggregates as a result of the sheer number of users. Because people generally want a user experience similar to Reddit, I think it’s inevitable that most user activity will be concentrated in one or two instances. It is probable that these instances follow in the footsteps of Reddit- the cycle repeats.
I actually think the Fediverse is in the beginning the process of fragmenting into siloed smaller, centralized instances. Beehaw, which is on the list of top instances, just blacklisted everyone from lemmy.world. Each of the three largest instances now are working to be a standalone replacement for Reddit and are in direct competition with each other. It is possible that this fragmentation and instability? of Lemmy instances will kill the viability of Federated Reddit altogether, but hopefully not.
These are my main takeaways from my three days on the Fediverse. I will stick around to see if the Fediverse can sustain itself after the end of the Reddit blackouts.
Especially since Lemmy doesn’t seem to support any sort of clustering or high availability setup at the moment - at least I don’t see any docs about it. Instances are currently expanding their capacity by vertically scaling (upgrading the server and adding more RAM, better CPU, more disk space, etc) but you reach a point where it’s better to horizontally scale (add extra servers) instead.
One large server is fine but the entire community goes down if it goes down. If you have multiple servers, you can just fail over to a different one.